Thermochimica Acta 676 (2019) 33–38 Contents lists available at ScienceDirect Thermochimica Acta journal homepage: www.elsevier.com/locate/tca Thermo-optical properties as complementary parameters for damage assessment of mineral oils aged under controlled conditions used in power transformers T A. Marín-Serranoa, J.A. Balderas-Lópezb, P.A. Calvaa, , A. Aranda-Péreza ⁎ a Instituto Politécnico Nacional, Unidad Profesional Interdisciplinaria en Ingeniería y Tecnologías Avanzadas (UPIITA), Av. Instituto Politécnico Nacional 2580, Gustavo A. Madero, Col. Barrio La Laguna Ticomán, 07340 CDMX, Mexico b Instituto Politécnico Nacional, Unidad Profesional Interdisciplinaria de Biotecnología (UPIBI), Av. Acueducto S/N, Col. Barrio la Laguna Ticomán, C. P. 07340, CDMX, Mexico A R TICL E INFO A BSTR A CT PACS: 84.70.+p 77.22.Jp Mineral oils are used in electric power transformers for their excellent dielectric properties. The continuous exposition to electrical discharges and thermal stresses may cause chemical adulteration in these substances which eventually could be cause of electrical catastrophic fails in the transformers. It is then important to evaluate periodically the quality of the oils to prevent major damages. The technique currently used for such purposes is gas chromatography, however its high inaccuracy iswell known. Chemical transformation of aged oils changes their thermal and optical properties and these could be used as alternative parameters for diagnostic purposes of electric power transformers. In this paper two configurations of photopyroelectric techniques were used for optical absorption coefficient (at 405 nm) and thermal diffusivity measurements of mineral oils aged up to 2300 h under controlled conditions. It was found that, at least up to this period of aging time, thermal diffusivity does not show significant differences, optical absorption coefficient at 405 nm, on the other hand, showed significant differences starting at 1000 h and then this optical parameter can potentially be used for diagnostic purposes. Keywords: Mineral oils Gas chromatography Photopyroelectric techniques Power transformers 1. Introduction Mineral oils are widely used as electrical insulators due to their high dielectric strength. Hence, these liquids are essential constituents in the isolation system of electric power transformers, which mainly consists of paper immersed in the mineral oil. Under working conditions this isolation system is subjected to high electrical efforts that cause chemical degradation of the paper and oil by hydrolysis and oxidation. Most frequent fails of electric power transformer are the result of the chemical transformation of these two constituents [1–3]. Chemical decomposition of the oil generates gases like hydrogen, methane, acetylene, ethylene and ethane, degradation of paper, on the other hand, generates carbon monoxide, carbon dioxide and furanic derivatives. On a difference of the oil, which can be changed, paper cannot be replaced whereby its damage is a major problem in power transformers and this is the most critical fail of this devices. Failures in transformers are the cause of considerable losses in the electrical industry [4], by which preventive maintenance is very important. Usual tests involve ⁎ detection of the gases which are produced during the operation of transformers and gas chromatography that is the standardized methodology in the electrical industry [5]. The central idea of using this analytical methodology is to study the gas development due to gradual chemical adulteration of the oils. More complete characterization involves other methodologies like neutralization number and polymerization degree, which revealed the chemical changes on the oil and paper, respectively. These techniques are expensive and time consuming whereby the search of other experimental criteria for the diagnosis of the quality of the insulation systems are relevant. Among other alternatives for the study of the degree of degradation of the oils are thermal analysis [6] and infrared spectroscopy [7] which involves qualitative analysis of new chemical species as results of the chemical degradation. Chemical adulteration of the oil due to aging may also changed thermal and optical properties at such extent that measurement of thermal conductivity or thermal diffusivity and optical absorption coefficients (at convenient wave-lengths) could be used as complementary parameters for better diagnostics of the state of electric Corresponding author. E-mail addresses: primoalbertocch@hotmail.com, pcalva@ipn.mx (P.A. Calva). https://doi.org/10.1016/j.tca.2019.03.025 Received 12 December 2018; Received in revised form 12 February 2019; Accepted 19 March 2019 Available online 20 March 2019 0040-6031/ © 2019 Elsevier B.V. All rights reserved. Thermochimica Acta 676 (2019) 33–38 A. Marín-Serrano, et al. Fig. 1. PPE experimental setup for thermal diffusivity measurements of liquids. Fig. 2. PPE experimental setup for optical absorption coefficient measurements of liquids. power transformers. Photothermal (PT) techniques have already been used for diagnostic purposes for oils [8–12], in this paper two configurations of photopyroelectric (PPE) technique [11,12] are used for the direct measurement of thermal diffusivity and optical absorption coefficient (at 405 nm) of aged oils up to 2300 h under controlled conditions. Some preliminary diagnostics about the quality of the oils are provided based on these thermo-physical properties supported at some extent on other more conventional studies as infrared spectroscopy. where D is a constant, independent of the sample´s thickness and βs is the optical absorption coefficient of the sample at the wave-length used for the analysis (a diode laser pointer at 405 nm in this case). The optical absorption coefficient of the sample is obtained directly as the slope of the linear fit of the PPE amplitude (in semi-log scale) as a function of the sample thickness. The constant behavior of the PPE phase is used in this case as experimental criterion for validate this measurement. [12] Systems for controlled degradation of the mineral oil To thermally deteriorate the oil, two identical aging tanks of 25 liters capacity were used, in which copper-paper coils were placed inside, representing a scale model of a power transformer of 75 MVA, keeping the copper/paper/oil relations. Fig. 3 is a photograph from de outside of the experimental system for the aging of the dielectric mineral oils at controlled thermal stress and Fig. 4 is a photograph of one coil inside one of the tanks. Fig. 5 shows the schematic diagram of the experimental arrangement. The controlled deterioration of the paper-oil system was carried out for a period of 2300 h at an average coil temperature, known as the hottest point or 140 °C hotspot, with oil circulation rate of 1 L per minute and with samplings at 300-hour intervals. 2. Experimental setups 2.1. Photopyroelectric setups for thermo-optical characterization The PT techniques described in this paper for the assessment of the thermal and optical properties of mineral oils have been published elsewhere for which just some relevant aspects of them are described here. Fig. 1 is a cross section of the PPE set up for thermal diffusivity measurements [11], it consists of a diode laser (Thorlabs, model L785P090, 90 mW CW power, λ = 785 nm) and a PVDF film as a pyroelectric sensor (0.0052 cm thickness). Laser radiation, modulated at 1 Hz, is absorbed on a thin silicon slab to generate thermal waves (temperature fluctuations inside the silicon slab) which travels through the liquid sample (thickness L), to reach the pyroelectric sensor. Under the limiting condition of thermally thick regime of the liquid sample, | s L| 1 (where s = (1 + i ) f s 3. Samples preparation , and αs is the thermal diffusivity of the sample), the pyroelectric signal, as a function of the sample´s thickness, is given by V (L) = Ce sL The mineral oil used for this work was the NYNAS NITRO 11GBXUS, which is not-inhibited naphthenic oil. To extract the moisture (1) where C is a complex expression, independent of L. The thermal diffusivity of the sample can be obtained from the amplitude and phase of Eq. (1) as the slope, M, of the linear fit of the PPE amplitude (in semilog scale) or phase, as a function of the sample thickness, by means of f the equation = 2 [11]. M Fig. 2 shows a cross section of the PPE setup for optical absorption coefficient measurements [12]. It is essentially the same as the one for thermal diffusivity measurements but, on a difference of it, the silicon slab is changed by an optical window as to allow light goes through the liquid sample. In the thermally thick regime for the liquid sample the pyroelectric signal is given in this case as V (L) = De sL Fig. 3. Experimental arrangement system for aging of the dielectric mineral oil. (2) 34 Thermochimica Acta 676 (2019) 33–38 A. Marín-Serrano, et al. Table 1 Dissolved gasses inside oil samples obtained from different power Transformers under real working conditions. Gas concentrations are given in ppm. Fig. 4. Coil inside the tank shown in Fig. 3. Gases Sample R1 3 years aging time Sample R2 4 years aging time Sample R3 5 years aging time H2 O2 N2 CH4 CO CO2 C2H4 C2H6 C2H2 0.000 27722.72 70183.87 1.392 8.667 341.630 1.402 0.804 0.000 0.000 33668.38 69446.05 0.763 4.747 147.649 0.058 0.000 0.000 0.000 20235.85 51321.02 0.000 4.185 59.499 0.000 0.000 0.000 degradation of the oil and possibly by the affectation of the cellulose. The dissolved gases that appear can be caused by thermal stresses due to overloads, electrical stresses due to corona discharges and / or the presence of electric arcs, moisture reactions with the air gap, among other factors. The state of the transformers depends not only on the time of service but also on how they are operated. Consequently in addition to the oil samples obtained with the procedure described above, three oil samples were obtained from different power transformers under real working conditions with operation times of 3, 4 and 5 years, and according to their gas chromatography, they reported the dissolved gases shown in Table 1. It is pertinent to anticipate that samples prepared with those obtained from real conditions are not comparable. Fig. 5. Experimental diagram of deterioration system of the paper-oil system. absorbed by the paper that covers the coils, these were entered into an autoclave system. The coils were subjected to a drying process at 100 °C (212 °F) under vacuum conditions for 120 h, after which it was possible to obtain a relative humidity in the paper of 0.75%. The coils were impregnated with low moisture mineral dielectric oil (15 ppm) and the thermocouples were connected to their respective terminals. Afterwards, the aging containers were sealed and filled with virgin mineral oil up to 90% of their capacity, keeping the terminals of the current bushing (passage-walls) immersed in the fluid. After filling, the air headspace was replaced with dry nitrogen until reaching 4 PSIG to prevent oil absorbing moisture from the air. After a leak-proof and stabilization time of 36 h under nitrogen pressure, the coil current was increased using an autotransformer for each one. It was necessary to increase the current intensity of the secondary at intervals of 10 amperes per minute until reaching the point at which the current stabilized and the temperature of both coils was maintained at 140 ± 2 °C. Once reached the temperature of 140 °C, the oil pumping device was put into operation to homogenize the temperature of the oil in its entirety. At the end of the 2300 h of aging, the coloration of the oil changed, from being virtually transparent to presenting a bright yellow color, which makes the oxidation evident as a consequence of the accelerated aging to which it was subjected. Fig. 6 shows the change in coloration of the oil. It is well known that in real operating conditions of electrical power transformers, it usually takes at least two years of continuous operation to be able to observe, through conventional techniques, significant deteriorations in the insulation systems, generated mainly by the 4. Results and analysis 4.1. UV vis characterization Oil samples aged under controlled conditions and three oil samples aged under real power transformer conditions were used for this study (Table 2, column 1). Fig. 7 shows the UV–vis spectra, for the three last samples (R1, R2 and R3). A strong absorption band around 400 nm is evident for all samples, showing the feasibility of using optical properties in the near UV as quantitative magnitudes for aging estimation of these kind of substances. It is important to note that even though this procedure could be carried out using the absorbance at a selected wave-length (400 nm for instance), this is not reliable since there is no certainty that this magnitude complies with Beer-Lambert law for light absorption. As it has been reported in the literature [12], the photopyroelectric technique reported to carry out the measurement of the optical absorption Table 2 Thermal diffusivity, αs, and optical absorption coefficient, βs (at 405 nm), measured by the PPE techniques described in this work, for the mineral oils samples aged under controlled (samples S1 to S7) and real working conditions (samples R1, R2 and R3). Fig. 6. Coloring of the oil at: A) 0 h; B) 2300 h of deterioration. 35 Oil sample βs (cm−1) (405 nm) αsx10−8 (m2/s) S0 (0 h) S1 (400 h) S2 (700 h) S3(1000 h) S4 (1300 h) S5 (1600 h) S6 (1900 h) S7 (2300 h) R1 (3 years) R2 (4 years) R3 (5 years) 0.16 0.12 0.17 0.11 0.23 0.25 0.34 0.41 0.66 2.22 3.92 6.69 6.63 6.56 6.58 6.68 6.63 6.53 6.62 6.65 7.20 7.43 ± ± ± ± ± ± ± ± ± ± ± 0.02 0.01 0.02 0.03 0.03 0.02 0.03 0.02 0.01 0.02 0.02 ± ± ± ± ± ± ± ± ± ± ± 0.06 0.06 0.05 0.06 0.05 0.05 0.06 0.06 0.04 0.07 0.06 Thermochimica Acta 676 (2019) 33–38 A. Marín-Serrano, et al. Fig. 9. Optical absorption coefficient, as a function of aging time, for the oil samples aged under controlled conditions in a range of 0–2300 h. Fig. 7. UV–vis spectra of the oils aged under real working conditions. It can be observed that it experiments appreciable changes starting from 1000 h of aging. These changes can be attached to the chemical degradation of the oils which involves conversion of chemical species inside the sample. It can be concluded that, at least for oils under this controlled conditions, appreciable chemical conversion is experimented up to 1000 h of aging. Qualitative determination of these chemical modifications to the oil molecules due to artificial aging was made using infrared spectroscopy. Fig. 10 shows the spectra obtained for the samples aging between 400 and 2300 h and the corresponding one for the sample without aging. The characteristics peaks for the functional groups in this figure were identified based on the nature of the oil molecule (oil for transformers of the brand NYNAS and naphthenic base). According to literature [13] the oil without aging is mainly composed of biphenol, naphthene and fluorene and this information was fed to the special software for the analysis (KnowItAll software) and using IR data tables. [14,15] The triplet between 2800 cm−1 and 3000 cm−1, corresponds to the CeH functional group as part of the fluorine. The doublet between 2310 cm−1 and 2380 cm−1, corresponds to the double bonds in the CO2 molecule and the peak located at 1730 cm−1 is related to the C]O bond. Fig. 10 shows that CO2 is a product of degradation of the oils since the corresponding peaks are evident starting at 1300 h of aging, time at which the corresponding peak for the C]O bond starts to decrease. The peak at 1315 cm−1 is related to the CeN bond; the presence of this nitrogen bond in the aged oils is attributed to chemical Fig. 8. Photopyroelectric phase, a, and amplitude (in semi-log scale), b, as a function of the sample thickness for an oil sample aged at 1600 h under controlled conditions. The continuous line in Fig. 8b corresponds to the linear fit for obtaining the coefficient of optical absorption of the sample at 405 nm. The measured value of this optical parameter is shown on the same plot. coefficient does not suffer from this deficiency since each measurement of this parameter is validated by the constant phase condition in the experiment. Taking into account this fact the optical absorption coefficients for all samples were obtained at 405 nm, this particular wavelength was chosen for this optical measurement since diode laser at this wave-length is commercially available. As an example Fig. 8 shows typical photopyroelectric signals, as function of the sample thickness, for a sample at 1600 h of controlled aging. The phase constant in Fig. 8a guarantee that the optical absorption coefficient, obtained as the slope of the linear fit in Fig. 8b, fulfill the Beer-Lambert model for light absorption. Table 2, column 2 summaries the corresponding results for all samples. Fig. 9 shows the behavior of this optical parameter as function of the time of aging. Fig. 10. Infrared spectroscopy of aged oil from 400 to 2300 h. 36 Thermochimica Acta 676 (2019) 33–38 A. Marín-Serrano, et al. Table 3 Effect of thermal aging on breakdown voltages. Aging time [h] Breakdown voltage [kV] 0 2400 33.94 20.88 absorption coefficient (Fig. 9), there is not an appreciable change on this thermal property over this period of aging time (1300 h) under these controlled conditions. However appreciable changes on this thermal property are evident for the samples obtained from transformer under real working conditions which where aging for more than 3 years, as it is evident by looking at the two last rows on Table 2. It is also evident that these changes are even larger for the optical absorption coefficient at 405 nm. This could be due to competitive behavior of the multiple new compounds which are generated during the aging process of the oil. Some of these compounds could promote and some other disturbs the heat transfer inside the oils. At relatively small periods of time (less than 1300 h) these differences in thermal transfer could be matched resulting in an overall null increment on thermal diffusivity; however this matching could be broken at larger periods of time, as it is shown in Table 2. The increase in the optical absorption coefficient at 405 nm, on the other hand, could be due to the increase in special chemical radical groups that promoted the UV radiation absorption and this fact increase monotonically with the aging time. To figure out on these assumptions larger periods of aging time under controlled conditions are required and it is part of another project. Finally, one of the dielectric properties with greater dependence on the physical-chemical characteristics of the oil is the dielectric breakdown voltage. This is indicative of the ability of the insulating liquid to withstand electrical stress. The presence of any of the mechanisms of deterioration of the oil, such as oxidation, is immediately reflected in its value of dielectric strength. For this work the dielectric breakdown voltages of the accelerating aging samples were determined with 0 h and 2400 h of deterioration, respectively. These values were obtained following the guidelines of the applicable regulations [16] and are shown in Table 3. Fig. 11. Photopyroelectric amplitude, as a function of the sample thickness for an oil sample aged at 0 h. The continuous line corresponds to the linear fit for obtaining the thermal diffusivity of the sample. The measured value of this thermal property is shown. Fig. 12. Thermal diffusivity, as a function of aging time, for the oil samples aged under controlled conditions in a range of 0–2300 h. combination with atmospheric nitrogen promoted by the electrical discharges (in the experimental container the air was substituted by dry nitrogen at a pressure of 4 PSI). It is interesting to note in Fig. 10 that the corresponding peak gradually increase, reaching its maximum around 1300 h of aging, and then decrease and disappear at 2300 h. Chemical degradation of the oil is evident from these IR spectra by looking at the peaks corresponding to CeN and C]O bonds. These facts correlate qualitatively well with the optical absorption measurements (Fig. 9) since at 1300 h the optical absorption coefficient at 405 nm also starts to increase. It is interesting to note that some peaks in Fig. 10 remain constant during the aging process; this is an indication that, at least under these controlled conditions, some molecular species inside the mineral oil remains unchanged. The type of gases dissolved in the volume of the oil and the associated chemical changes depend on the thermal and electrical stresses to which it is subjected. These global changes eventually will play on detectable changes on its thermal properties. To figure out about this effect pyroelectric technique was used for the measurement of this thermal property for all samples. Fig. 11 shows the photopyroelectric amplitude, as a function of the sample thickness, for the oil sample aged at 0 h for the goal of illustrate the analytical procedure for obtaining the thermal diffusivity of the sample for all samples. Table 2, column 3, summarized the thermal diffusivity values for all of the samples studied in this work. Thermal diffusivity as a function of the time of aging is shown in Fig. 12. It is evident from this figure that, at a difference of the optical 5. Conclusions For the first time photopyroelectric techniques were used for obtaining optical absorption coefficient and thermal diffusivity measurements of aged mineral oil from power transformers. It was found that optical absorption coefficient, at 405 nm, of the aged oil can be used as a complementary parameter to diagnose the state of the insulations. The optical photopyroelectric analysis described in this work could be easily carried out at other convenient wave-lengths (in the UV-region) as to provide of more optical information and relate it with the different chemical species which are formed as result of the degradation process. Acknowledgements This work was supported by the Instituto Politécnico Nacional (Grant 2016812) and the Instituto Nacional de Electricidad y Energías Limpias, of México. References [1] Lars E. Lundgaard, Walter Hansen, Dag Linhjell, Terence J. Painter, IEEE Trans. Power Delivery 19 (1) (2004) 230. [2] Marit-Helen Glomm Ese, Knut B. Liland, Lars E. Lundgaard, IEEE Trans. Dielectr. Electr. Insul. 17 (3) (2010) 939. [3] Shayan Tariq Jan, Raheel Afzal, Akif Zia Khan, Int. Conf. Data Mining, Civil and Mechanical Engineering (ICDMCME’2015), (2015), p. 49. [4] CIGRE Technical Brochure, (2004), p. 248. [5] IEEE Guide for the Interpretation of Gases Generated in Oil-Immersed Transformers, (2008), p. 3. 37 Thermochimica Acta 676 (2019) 33–38 A. Marín-Serrano, et al. [6] Marina A. Tsili, et al., Int. J. Therm. Sci. 53 (2012) 118. [7] A. Abu-Siada, P. Lai Sin, S. Islam, Power Systems Conference and Exposition (PSCE 2009), (2009), p. 1. [8] J.A. Balderas-López, A. Mandelis, Self-consistent photothermal techniques: application for measuring thermal diffusivity in vegetable oils, Rev. Sci. Instrum. 74 (1) (2003) 700. [9] Dane Bicanic, Otto Dнka, Svjetlana Luterotti, Andreas Bohren, Mateja Šikovec, Bebvan Veldhuizen, Otto Berkessy, Mihai Chirtoc, Mladen Franko, Gabor Szabo, Marcus Sigrist, Assessing the extent of oxidation in thermally stressed vegetable oils. Part I: Optical characterization by photothermal and some conventional physical methods, Anal. Sci. 17 (2001) s547. [10] M. Streza, D. Dadarlat, C. Socaciu, C. Bele, F. Dulf, V. Simon, Food Biophys. 4 (2009) 147. [11] J.A. Balderas-López, M.R. Jaime-Fonseca, G. Gálvez-Coyt, A. Muñoz-Diosdado, J. Díaz-Reyes, Generalized photopyroelectric setup for thermal-diffusivity measurements of liquids, Int. J. Thermophys. 36 (2015) 857–861. [12] J.A. Balderas-López, Generalized 1D photopyroelectric technique for optical and termal characterization of liquids, Meas. Sci. Technol. 23 (2012) 10 065501. [13] Isaac R. Kaplan, Jimmy Rasco, Shan-Tan Lu, Environ. Forensics 11 (1–2) (2010) 117. [14] F. Govaert, M. Bernard, Forensic Sci. Int. 140 (2004) 61. [15] G. Socrates, JOHN Wiley & Sons ltd (Ed.), Infrared and Raman Characteristic Group Frecuencies Tables and Chart, The University of West London, Middlesex UK, 2005. [16] ASTM Standard Test Method for Dielectric Breakdown Voltage of Insulating Liquids Using Disk Electrodes, ASTMD877/D877M-13, ASTM INTERNATIONAL, WEST CONSHOHOCKEN, PA, USA, 2013. 38